With this, alias defs become singletons based on the def they alias and the
type and offset of the alias. Thus, the removal of the free_def call in
emit.c.
alias_def now always creates an offset def (though the usual case has an
offset of 0). The if the alias escapes the bounds of the base def, an
internal error will be generated.
It really doesn't seem wise to allow the compiler to do so as it would
overwrite unrelated defs. The only time such a thing is valid is the return
statement (silly vm design), and that's read-only.
Also remove the extern for current_storage as it belongs in shared.h.
I'm not satisfied with the documentation for initialize_def, but it will do
for now. I probably have to rewrite the thing as it's a bit of a beast.
With the intoduction of the statement type enum came a prefix clash. As
"st" makes sense for "statement type", I decided that "storage class"
should be "sc". Although there haven't been any problems as of yet, I
decided it would be a good idea to clean up the clash now. It also helps
avoid confusion (I was a bit surprised after working with st_assign etc to
be reminded of st_extern etc).
It doesn't quite work yet, but...
It has proven necessary to know what type .return has at any point in the
function. The segfault in ctf is caused by the return statement added to
the end of the void function messing with the expr pointer stored in the
daglabel for .return. While this is actually by design (though the
statement really should have a valid expr pointer rather than), it actually
highlights a bigger problem: there's no stable knowledge of the current
type of .return. This is not a problem in expression statements as the
dagnodes for expression statements store the desired types of all operands.
However, when assigning from .return to attached variables in a leaf node,
the type of .return is not stored anywhere but the expression last
accessing .return.
Now information like dags or live variables are dumped separately, and the
live variable information replaces the flow node in the diagram (like dags
have recently).
They really should have been in statements.[ch] in the first place
(actually, they sort of were: is_goto etc, so some redundant code has been
removed, too).
The evil comment is not just "pragmas are bad, ok?", but switching between
advanced, extended and tradtitional modes when compiling truly is evil and
not guaranteed to work. However, I needed it to make building test cases
easier (it's mostly ok to go from advanced to extended or tradtional, but
going the other way will probably cause all sorts of fun).
In the process, opcode_init now copies the opcode table data rather than
modifying it.
After running across a question about lists of animation frames and states,
I decided giving qfcc the ability to generate such lists might be a nice
distraction from the optimizer :) Works for both progs.src and separate
compilation. No frame file is generated if no macros have been created.
It is necessary to know if a def is a function parameter so it can be
treated as initialized by the flow analyzer. The support for the flag in
object files is, at this stage, purely for debugging purposes.
.return and .param_N are not classed as global variables for data flow
analysis. .return is taken care of by return statements, and .param_N by
call statements.
With this, the menus work up to attempting to load the menu plist.
Something is corrupting zmalloc's blocks.
With temp types changing and temps being reused within the one instruction,
the def type is no longer usable for selecting the opcode. However, the
operand types are stable and more correct.
Nicely, the need for dag_gencode to recurse seems to have been removed.
At least for a simple case, correct code is generated :)
switch.r:49: case 1: *to = *from++;
003b loadbi.i *(from + 0), .tmp10
003c add.i from, .imm, from
003d storep.i .tmp10, *to
It doesn't make any difference yet, but that's because I need to add extra
edges indicating iter-node dependencies. However, the sort does seem to
work for its limited input.
While things are quite broken now (very incorrect code is being generated),
the dag is much easier to work with. The dag is now stored in an array of
nodes (the children pointers are still used for dagnode operands), and sets
are used for marking node parents, attached identifiers and (when done,
extra edges).
Instead of storing the generating statement in the dagnode, the generating
expression is stored in the daglabel. The daglabel's expression pointer is
updated each time the label is attached to a node. Now I know why debugging
optimized code can be... interesting.
It now seems to generate correct code for each node. However, node order is
still incorrect in places (foo++ is being generated as ++foo). quattest.r
actually executes and produces the right output :)
flow_analyze_statement uses the statement type to quickly determin which
operands are inputs and which are outputs. It takes (optional) sets for
used variables, defined variables and killed variables (only partially
working, but I don't actually use kill sets yet). It also takes an optional
array for storing the operands: index 0 is the output, 1-3 are the inputs.
flow_analyze_statement clears any given sets on entry.
Live variable analysis now uses the sets rather than individual vars. Much
cleaner code :).
Dags are completely broken.
The types are expression, assignment, pointer assignment (ie, write to a
dereferenced pointer), move (special case of pointer assignment), state,
function call/return, and flow control. With this classification, it will
be easier (less code:) to determine which operands are inputs and which are
outputs.
Surprisingly, I don't yet have to "throw one out", but things are still
problematic: rcall1 is getting two arguments, goto and return get lost,
rcall2 got an old temp rather than the value it was supposed to, but
progress :)
First, it turns out using daglabels wasn't such a workable plan (due to
labels being flushed every sblock). Instead, flowvars are used. Each actual
variable (whether normal or temp) has a pointer to the flowvar attached to
that variable.
For each variable, the statements that use or define the variable are
recorded in the appropriate set attached to each (flow)variable.
The flow graph nodes are now properly separated from the graph, and edge
information is stored in the graph struct. This actually made for much
cleaner code (partly thanks to the use of sets and set iterators).
Flow graph reduction has been (temporarily) ripped out as the entire
approach was wrong. There was also a bug in that I didn't really understand
the dragon book about selecting nodes and thus messed things up. The
depth-first search tree "fixed" the problem, but was really the wrong
solution (sledge hammer :P).
Also, now that I understand that dot's directed graphs must be acyclic, I
now have much better control over the graphs (back edges need to be
flipped).
The reduction is performed itteratively until the graph is irreducible, but
such that each reduction wraps the previous graph. Unfortunately, due
depth-first searching not being implemented, graphs that should be reduced
(ie, those with natural loops).
set_first() now returns a pointer to a setstate_t struct that holds the
state necessary for scanning a set. set_next() will automatically delete
the state block when the end of the set is reached. set_delstate() is also
provided to allow early termination of the scan.
They're now dot_sblock.c and print_sblock. The new names both better
reflect their purpose and free up "flow" for outputting the real flow
analysis graphs.
Much of the data recently added to sblock_t has been moved to flownode_t.
No graph reduction is carried out yet, but the initial (innermost level)
graph has been built.
All internal structs now have "proper" names, and fit the naming convention
(eg, obj_module (like objective-c's types, but obj instead of objc). Some
redundant types got removed (holdovers from before proper struct tag
handling).
Also, it has proven to be unnecessary to build internal classes, so
make_class and make_class_struct are gone, too.
When encoding a type to a qfo file, the type's encoding string is written
and thus needs to be valid prior to actually doing the encoding. The
problem occurs mostly in self-referential structs (particularly, obj_class)
because the struct is being encoded prior to the pointer to the struct.
Type names are cleaned up, as is the creation. Also, the class pointer in
the type encoding now gets emitted. However, Still need to actually create
_OBJ_CLASS_Class and fix the type encoding reloc handling in the linker.
Since gnu bison and flex are required anyway, no harm in using their api
prefix options. Now, qfcc can compile both QC/Ruamoko and Pascal files
(Pascal is (currently?) NOT supported in progs.src mode), selecting the
language based on the extension: .r, .qc and .c select QC/Ruamoko, .pas and
.p select Pascal, while anything else is treated as an object file (as
before).
The output can be controlled via --block-dot (not yet documented). The
files a named <sourcefile>.<function>.<stage>.dot. Currently, stage will be
one of "initial" (after expression to statement conversion), "thread"
(after jump threading), "dead" (after dead block removal), "final" (final
state before actual code emission).
It is inteded for flagging buggy conditions in the compiler, particularly
after having fixed the original bug (in case something comes back from the
dead).
Despair has things locked down such that running qfcc during a build fails
due to lack of read access to /usr/local/lib. This is actually a good
thing as accidentally hitting old includes/libs (when a file gets deleted
in the tree) hides bugs. Thus, --no-default-paths to turn off default
search paths.
Statement operands throw away the high level type information, so store
type size in the operand and use this size for allocating space for temps
rather than using the low-level type.
"vector-components" in code options controls this feature. The default is
off for advanced code and on for traditional code. Disabling
vector-components prevents the comonent names polluting the namespace and
reduces the number of globals needed for vector fields if the components
of that field are never used.
There is much breakage, but qfcc now produces a progs.dat from either
progs.src or object files. Better yet, the progs.src result is passed
through the linker, removing much duplicate code.
Structures etc now never encode the fields as well (I might revert that
at some time). Type parsing has been removed: it has proven unnecessary
and overly complicated, and the new qfo encoding should be more useful.
Access to struct fields in near data can be done using only one operand,
but offset relocs need to be used. However, as not all defs want offset
relocs, a flag has been added to the def struct.
The space is meaningless for op_* relocations as they are always in the
code space, but def_* relocations need to know which space holds the
location to be adjusted.
find_type now operates recursively (depth-first) so built up typechains
work as expected.
@overload is treated as a specifier (directly as a storage class, similaar
to typedef).
Separate compilation will take fixing object files.
The generated code is broken due to various relocation fixups being
broken, and float immediates seem to be badly broken.
Debug information is broken too.
The first function seems to work fine, but there's a problems with the
scope of params causing params to get their knickers in a twist (tangled
linked list).
Since there is now a proper symbol table, defs are now just references to
memory locations and the symbol table takes care of duplicates.
Also, start using far data for ObjQC structures.
The qfo functions have been stubbed out until I figure out what to do with
object files in the new scheme.
That which isn't rewritten is horribly broken. However, this does include a
nice mechanism for building QC structs for emitting data.
emit.c has been gutted in the spirit of "throw one away".
There is much work to be done to get even variables emitted, let alone
code. Things should be a little more fun from here on.
Use "@reference ClassName;" or "@reference ClassName(CategoryName);" to
create a refence to the class or category, forcing the defining object file
to be linked into the program when the object file is part of a library.
define default include and lib paths for qfcc
ruamoko:
{cl_menu,game}/Makefile.am:
conform to the new qfcc library linking rules
lib/Makefile.am:
install the libs to ${prefix}/lib/ruamoko
qfcc:
linker.[ch]:
support path searching for -llib and make linker_add_lib search for
libfoo.a for -lfoo in the paths, or just a single dir search for the
libname otherwise.
options.c:
support -L libpath and setup the default include and lib paths
also change most strdup calls to save_string
qfcc.c:
check for foo.a as well as -lfoo when deciding whether the file is
an object file or lib file.
call obj_find_message with super instead of class when doing a super
lookup
client_menu.[ch]:
use the InputLine object instead of directly using the api
InputLine.r:
builint functions are = #0, not = 0 :P
function.[ch]:
provide copy_params
method.[ch]
provide copy_keywordargs
expr.c:
call inc_users for the parm temp if it's a bind expression
revers a copy of the selector when building the selector name so the
selector doesn't get mangled
qc-parse.y:
catch erronous func = const inintializers (should be func = #const)
start working on the linker.
class.[ch]:
redo class defs so the pointer works (needs relocs still)
obj_file.h:
add prototype for read_obj_file, QFO_* data access macros and include
pr_debug.h
type.[ch]:
separate type system initialisation and recording of the standard types
so find_type works properly in multiple compilation
def.c:
don't mark static defs as initialized
expr.c:
proper class pointer def stuff
immediate.c:
clean out dead vars/code
obj_file.c:
allocate space for the line info and bail if the file can't be opened.
qfcc.c:
register the standard types for each compile pass and start linking the
files
new va function: nva which returns a strduped buffer
expr.c options.c:
use nva instead of strdup (va (...
struct.c type.c:
make type encoding work properly for structs
*type_method to type_Method
emit.c:
support casting between pointers
expr.c:
support casting between pointers
method.c:
correct the type for _cmd
type_method to type_Method.aux_type
qc-lex.l:
Method type is a poniter to a method
qc-parse.y:
support , args to messages (not fully implemented yet)
type.c:
*type_method to type_Method and make type_Method a pointer to a method
keywords (quaternion integer function for break continue switch case default
NIL struct enum typedef) and converts some errors to warnings (assignment to
constants, insufficient function arguments, return; from non-void function,
anal function `pointer' type checks)
o add a "freed" marker to def_t to prevent double freeing of
temp def offsets
emit.c:
o break out the bind code into emit_bind_expr (doesn't /really/
emit code, but still:)
o make bind work with disparate types (forces def_t->freed 1)
pr_def.c:
o mark array pointers as initialized and constant.
o don't free the offset for already freed temp defs
pr_imm.c
o fix a bug in uinteger support
o support disparate types for immediates
switch:
o fix a bug where gt.i was being used instead of gt.ui
o remove some debug code
o add OP_JUMPB
o OP_JUMPB renumberd some opcodes, so up PROG_VERSION
pr_edict.c:
o make the version error reporting more informative
pr_exec.c:
o implement OP_JUMPB (goto *(ptr + index))
pr_opcode.c: (libs/gamecode/engine)
o add OP_JUMPB to the table
expr.h:
o ex_uinteger support
o some const correctness
o prototype new_label_expr
qfcc.h:
o uinteger ussport
o add pointers for op_jump and op_jumpb
o prototype PR_GetArray
emit.c:
o general uinteger support
o new reference/reloc type 3: absolute statement address
o jumpb support (binary goto)
expr.c:
o uinteger support
o break the label name creation out of new_label_expr into
new_label_name
o some const correctness
pr_def.c:
o add PR_GetArray to allocate an array in global space
o factor out some code common to PR_GetDef and PR_GetArray that would
otherwise be duplicated
pr_imm.c:
o some const correctness
o uinteger support
pr_lex.c:
o uinteger support
pr_opcode.c: (tools/qfcc/source)
o support jump and jumpb
switch.c:
o rewrite the binary search code to support ranges.
o add ev_uniteger to the types enum
o add opcodes for ifbe, ifb, ifae, ifa, jump, lt.ui, gt.ui, le.ui, ge.ui
progs.h:
o add uinteger accessors
pr_exec.c:
o implement ifbe, ifb, ifae, ifa, jump, lt.ui, gt.ui, le.ui, ge.ui
pr_opcode.c:
o add opcodes for ifbe, ifb, ifae, ifa, jump, lt.ui, gt.ui, le.ui, ge.ui
expr.h:
o prototype inc_users
qfcc.h:
o add externs for op_ifbe, op_ifb, op_ifae and op_ifa
emit.c:
o don't bother emiting an assignment to a temp def that's only used once
(ie, it's never read, only written to)
o support the new if* instructions
expr.c:
o support the new if* insructions
o dectect expression loops in append_expr
o support unsigned integers
o re-work temp def usage counting
pr_def.c
o debugging for temp def usage counts
pr_opcode.c:
o support the new if* instructions
qc-parse.y:
o provide defines for IFBE IFB IFAE IFA
switch.c:
o do binary searches for strings, floats and ints if there are more than
8 cases in a switch. Strings need more testing.
get_type and extract_type prototypes and add extern for type_names
emit.c:
use extract_type instead of get_type
expr.c:
get_type now returns the type pointer rather than the qc type enum.
extract_type uses get_type to return the qc type enum.
make type_names global
use get_type instead of ex_expr/ex_def to cover more cases
clean up function call/return processing as a result of get_type
pr_opcode.c:
use opcode_priority_type_table_ab when var_c is null and a minor
cleanup in PR_Opcode_Find
*/CRITICAL/* implication: DO NOT /EVER/ RELY ON EVALUATION ORDER. This
isn't /really/ that bad: evaluation order will be consistent for similar
expressions, but for (eg) foo() + bar()*2, bar will be called first.
use reference counting for temp defs and call PR_FreeTempDefs at the end
of emit_sub_expr as well as emit_expr.
fix a (relatively harmless) bug in type processesing of foo.bar =
another builtin by name, and returns it.
Soon I'll change all our new builtins to by allocated dynamically, as
well as changing the number checkfunction uses, and happily break
everything that uses them :D
integer constants and float function args/return values.
pr_comp.h:
o add the integer opcodes to pr_opcode_e
pr_edict.c:
o add "quaternion" and "integer" to type_name[]
o support quatnernion and integers types when printing values
o support the integer opcodes when bounds checking
pr_exec.c
o enable the integer opcodes
pr_opcode:
o add the integer opcodes to the opcode table
o logical operators all result in an integer rather than a value
expr.h:
o rename int_val to integer_val
qfcc.h:
o kill another magic number
expr.c:
o move the opcode to string conversion out of type_mismatch and into
get_op_string
o rename int_val to integer_val
o general integer type support.
o generate an internal comipiler error for null opcodes rather than
segging.
pr_imm.c:
o rename int_val to integer_val
o support integer constants, converting to float when needed.
pr_lex.c:
o magic number death and support quaternions and integers in type_size[]
qc-lex.l
o rename int_val to integer_val
o support quaternion and integer type keywords
qc-parse.y:
o rename int_val to integer_val
o use binary_expr instead of new_binary_expr for local initialized
variables
builtins.c:
o rename int_val to integer_val
o fix most (all?) of the INT related FIXMEs
defs.qc:
o use integer instead of float where it makes sense
main.c:
o read_result is now integer rather than float
main.qc:
o float -> integer where appropriate
o new test for int const to float arg
nuke the ex_statement and estatement_[st] stuff
add label_expr prototype
expr.c:
ex_statement nukage
correct new_expr's decl
add label_expr to ease label creation
don't crash when printing a null expression (bare return)
qc-parse.y:
estatement_t nukage
statement statements and statement_block are type expr
generate `expressions' for statements
a full parse tree for each function is now generated. there are several
special expression opcodes for statements:
d done \
r return -> unary: expression to return or null
i if binary: evaluated expression, destination label
n ifnot binary: evaluated expression, destination label
c call binary: function def, args (expr list, rev order)
s state binary: frame const, function def
g goto unary: destination label
l label unary: label number
in a top level expression, l (label) defines the label, otherwise it is a
reference.
add prototype for print_expr
expr.c:
add print_expr
correct string accessors
currect the result type for unary operators
qc-lex.l:
correct string, vector and quaternion parsing
qc-parse.y:
precedence corrections and more function scope work
CustomTF gets through the parsing again.
rearrange def_t a little and add def_next (leaving next free for other
uses)
pr_def.c:
use def_next instead of next to link /all/ of the named defs
qfcc.c:
ditto
define PROG_ID_VERSION as 6 and redefine PROG_VERSION as 0x00fff001
(0.fff.001) for the new qc features.
pr_edict.c:
support version 6 and version 0.fff.001 progs
qfcc.h:
add version field to options_t
add min_version field to opcode_t
pr_opcode.c:
set the minumum version of each opcode (gee, that table is getting ugly)
filter out opcodes with too high a min_version when initializing the opcode
hash tables.
qfcc.c:
update help output.
accept --id to limit code generation to id compatable (ver 6) progs.
default progs generation to 0.fff.001
Detect assignments to initialized globals and give an error, unless the --cow
(copy on write) option is given, and then allocate a new global for the var,
clear its initialized flag.
Relocate all globals.
termporary variables sit in one pool of memory (at the end of the globals)
thus drasticly reducing globals requirements. This works because the whole
lot is declared to be in the function's local variable space which is copied
to the locals stack in the progs engine.
to get the opcode from the table record rather than the location within the
table (ewww). gives a nice speed boost /and/ makes the opcode table easier to
maintain.
add OP_ADD_S. WARNING!!! this /will/ move.
progs.h:
add prototype for PR_PrintStatement
pr_edict.c:
add OP_ADD_S support in the progs checker
pr_exec.c:
implement OP_ADD_S
tools/qfcc/include/.gitignore:
add config.h.in
qfcc.h:
nuke PR_NameImmediate and change PR_ParseImmediate's prototype (see
pr_imm.c)
pr_comp.c:
add ADD_S, adjust for PR_ParseImmediate's prototype, make
PR_ParseExpression work with non-sequential opcodes (slow, will work on
that next). Fix up initialised global parsing.
pr_imm.c:
nuke PR_NameImmediate. didn't work well and wasn't such a good idea anyway.
PR_ParseImmediate now accepts a def_t * arg. if null, will allocate a
new global def, otherwise it will initialize the def passed in.
qwaq/main.c:
sports some debugging code (dumps info about the progs it's running)
qwaq/main.qc:
better ADD_S testing
